Shrinivas M. Soman, Unigraphics
Presented June 16, 2000 at the ADCATS 2000 Conference

Statistical tolerance analysis of flexible assemblies can predict the probable range of assembly forces and distortions required to assemble warped or misaligned parts. The resulting distortion and stress throughout the assembly are affected by the amplitude and wavelength of the surface waviness of the mating surfaces. Therefore, the surface variations must be characterized statistically. A new method is presented, called the Functional Surface Characterization (FSC) method, to characterize a population of surfaces statistically.

The FSC method classifies the surface variations as random and non-random. The method separates the two variations by statistical analysis and uses frequency spectrum methods to characterize the random variations. The outputs of this method, the mean surface and average autospectrum, can be used with a statistical finite element analysis model to predict assembly forces and distortions. The autospectrum is used to classify the spatial frequencies in three domains namely, low, medium and high, based on distinctly different effects. The method is demonstrated using real surface data acquired from sample flexible parts to evaluate the spectrum model.

Bio

• BS ME - College of Engineering Poona, India ’92
• MS ME – BYU Aug. ‘99. Emphasis - CAD, Tolerance Analysis
• Worked as Design Engineer with a diesel engine manufacturer for 3.5 years.
• Worked on tool design of automotive castings.
• Currently working at Development Center of Unigraphics Solutions, California.
• Designation - Systems Engineer, Geometric Tolerancing Group
  Duties - Software Design, Implementation and Testing